Making Artificial Intelligence Omnipresent

20 Jun

There are several contenders in the AI and IoT fields. How can startups break through? Kneron successfully carved its way into the market. In 2015, Albert Liu, founder and CEO of Kneron, had seen the increase in hardware computing performance. That is why he incorporated AIoT’s intelligent edge computing into products. He began working with big companies to collect the necessary data for training AI, giving his startup a head start in the AI field. Kneron has partnered with Alibaba, Tencent, Baidu, Sogou, Arm Holdings, Qualcomm, TSMC, and Foxconn. They have helped Kneron catch the attention of investors, raising US$ 33 million since the start of its business. Kneron’s investors include Horizons Ventures, Alibaba Entrepreneurs Fund, CDIB, Himax, Qualcomm, ThunderSoft, Sequoia Capital’s sub-fund Cloudatlas, and CYZone.

A Complete Solution, Making AI Omnipresent

“Kneron’s vision is to make artificial intelligence present everywhere! The scenes in science fiction movies have become a reality. Whether it is images, voice, radio waves, or sound, with artificial intelligence, you can get a meaningful output for all of them,” said Roger Liu, the COO of Kneron.

Therefore, from image recognition software, the terminal device- specific AI processor “NPU IP Neural Network Processor” to the  AI SoC hardware, Kneron has created a complete software and hardware integration solution. The IoT is made even smarter with the AI solution, so that AI analysis of information like images and sound collected from IoT sensors enhances IoT device judgment and convenience of living.

Kneron provides various SoCs for different applications in the  AI ecosystem. Kneron has created solutions not only for high- performance edge computing but lightweight and low-power consumption, providing a much more efficient computing architecture. Roger Liu pointed out, “We can be very efficient for each critical step. With our total solutions, Kneron makes its way into the AI ecosystem cycle.”

Kneron’s solutions support three major applications. The first application is a mobile phone that supports AI smart scene detection, 3D modeling, or biometric camera lens for facial recognition or door security.

The second application is a smart home. The camera is used to detect body features; for example, when a child crawls to the kitchen from the living room or is near the furnace, an alert will be sent to the parents. If an elderly person falls, a notification will also be sent. The third application is security and surveillance, like notification of vehicle parking violations.

In terms of Kneron’s competitive advantage, Roger Liu mentioned there are currently two types of AI chips. One of them is cloud computing provided by companies like NVIDIA and Intel; the other is Kneron’s terminal unit computing, which is edge computing. This is how Roger Liu describes Kneron’s AI terminal computing: “Instead of the cloud, our terminal unit is the cerebrum and the application is the cerebellum. This gives the terminal unit automatic smart judgment.”

Kneron’s NPU IP Neural Network Processor Series is a terminal unit-specific AI processor. It is low power and small but has extremely high-performance computing and excellent power usage effectiveness.

The Three Stages of Entrepreneurship: Creating the AI chip and Solutions

Great dreams are not realized in one step. Roger Liu pointed out that the market for AI applications is huge. Startups face the challenge of how to define the technology, software, IP, and solutions, which need to be laid out in stages for products to be established. “We are one of the few startups that already have established products and revenue-generating AI solutions.”

During the first stage of starting up, Kneron found a way into the market with neural network image recognition software, including facial recognition, body and gesture recognition, and item and scene recognition. Kneron made the company profitable through commercial software.

In the second stage in 2017, Kneron stepped up AI intellectual property core production and launched the “NPU IP Neural Network Processor,” a terminal-unit specific AI processor. Albert Liu said that at the time, Kneron was working with investment companies and received recognition from international companies through the IP licensing model. Kneron carried out co-promotion agreements with large intellectual property core companies Synopsys and Cadence to establish their advanced technical products.

In the third stage, Kneron began planning AI SoC after financing was in place. In 2018, Kneron developed a total solution and will launch the first AI SoC product in the second quarter of 2019, focusing on the smart home market. In addition to working on the integration of AI SoC and industrial computers with Aaeon, Kneron will co-develop a terminal AI 3D sensor solution with Etron Technology.

This year, Kneron will expand recruitment and carry out a Series B funding round to find the next pivot for development. Kneron will also actively develop markets in Taiwan, Europe, US, and China, and ultimately, aim for IPO.

Roger Liu, COO of Kneron, believes the AI applications market is very large, but startups must define the technology, software, IP, and solutions, and lay them out in stages in order to establish products.

(Image credit: Shutterstock)

Facebook Comments